Takeoff from runway
Getting your Trinity Audio player ready...
4 min read

A major purpose of gathering accident statistics is to assess where safety resources should be allocated. In other words, are pilots being trained on the right things? Are safety messages targeting the right things? Are researchers’ efforts addressing problems with the most likely payback? And, as the title states, are “impossible turns” worth the effort that is being expended upon their study?

Similarly, there are often statements made that takeoff and go around accidents are underplayed compared to approach and landing accidents. That’s another study for another time, but this study did suggest how much takeoff and go around accidents compare with overall accident rates.

Takeoff from runway

Many takeoff accidents can be categorized as “failure to achieve sustained flight.”

The data set for this study was all NTSB accidents in 2018 involving single engine, fixed wing, piston airplanes, Part 91 operations, excluding IMC, student pilots, instruction, seaplanes, cruise conditions, landings, and night. The exclusions were done with an arduous, iterative, and painful combination of database search specification, keyword exclusions, and manual sorting. For example, “float” was used to eliminate float plane events till it was discovered that this keyword also excluded power losses due to carburetor floats. There were other gotchas, equally obscure. 1,078 events were found with the SQL search, trimmed to 137 with manual exclusions.

Although the analysis undoubtedly contains subjective errors, sorting errors, and poor data quality factors, it is not believed that any of those errors affect the conclusions. Remember that the purpose of this study was not to generate perfect statistics. The purpose of this study was to address the specific question in the title, so, for once, close enough was good enough.

As was expected, conventional taxonomies did not adequately describe the phenomena encountered in the data. Go Around events and Takeoff events shared some of the same taxonomy.

The terms used in this analysis were:

  • Failure To Achieve Sustained Flight: This major category meant that after power was applied, the airplane did not achieve flight beyond obstacles surrounding the airport, such as trees and power lines. Included in this category are aborted takeoffs, failure to avoid hitting an obstacle to the side of the runway either on the ground or in the air, inadequate power for the airplane to even take off, and a number of unlikely special cases. While this kind of catch-all category would be unacceptable in many analyses, it is appropriate here to differentiate situations in which an “impossible turn” was unlikely to be attempted from situations where an “impossible turn” was a possibility.
  • Forced Landing: The pilot landed the airplane off the airport property. In some of these cases, the “impossible turn” was not attempted because of good judgement.
  • Return To Runway Failed: The “impossible turn” was attempted and failed. Most severe injuries in these four events were one each None, Minor, Serious, and Fatal.
  • Return To Runway Succeeded: What we all cheer for, the impossible turn succeeded, even though the plane was damaged.
  • Chute Deployed: There were two of these, one in a Cirrus and the other in a rare LSA.

The NTSB reports had a marvelous term, “Abnormal Ground Contact,” used in the context of landings. That term was not relevant in this study, however, because the Go Around statistics did not consider the cause of the Go Around.

88 Takeoff Events

The majority of the Takeoff Events, 73%, were Failure To Achieve Sustained Flight, three fatal. Forced landings were 22%, none fatal. There were four Return To Runway Failed, only one fatal, two Return To Runway Succeeded, and two chute deployments. There may be additional Returns To Runway Succeeded that were not reported to the NTSB because there was no damage.

10 Go Around, No Touchdown

These were all Failure To Achieve Sustained Flight except for one Forced Landing. There were no fatalities, and only one Serious injury. There were undoubtedly many go arounds that were not accident events and hence, not reported.

34 Go Around With Touchdown

Almost all of these were Failure To Achieve Sustained Flight, with no fatalities and only two serious injury accidents. There was one Forced Landing and two cases where the damaged airplane completed the go around and flew to a landing.

5 Touch and Go

Again, only one Forced Landing and the rest were Failure To Achieve Sustained Flight. There were no injuries.

Overall Injury Statistics

These statistics are not necessarily precise but are representative of overall injury statistics, with the number of events in all of the events downloaded (including cruise and landings) in the center column, and the number of takeoff / go around events in the right column.

Accident stats

The rates of accident severity are comparable between TO/GA and all accidents, except for fatalities, which shows an obvious difference.

Concluding Thoughts

This limited data set and this admittedly less than robust analysis indicate that pilots are likely handling the “impossible turn” scenario properly, more often than not.

Similarly, the takeoff and go around accidents appear to generate only a very small proportion of fatal accidents.

A surprise in the narratives is the number of events in which the pilot flew down the runway with the stall horn blaring before the accident. The unanticipated conclusion is that pilot training should include flight near the stall and possibly full stalls before a pilot encounters a situation which could result in Failure To Achieve Sustained Flight.

Ed Wischmeyer
23 replies
  1. John Comley
    John Comley says:

    Let’s face it, under the pressure of an EFATO, a lot can go wrong, including turning with too low airspeed, in which a stall and spin-in becomes a distinct possibility, in most cases resulting in a fatal crash.
    And it can happen to ANYONE, including highly skilled and experienced airmen.
    Look at the recent ‘Reno-racer’ Sherman Smoot turn-back that resulted in his crash RIP.
    By having a Vdmms bug on your airspeed indicator, you can avoid such a stall, even when in a power-off turn.

    • Byron Huff
      Byron Huff says:

      I think that an AOA gauge would be 0f much more use than a bug to tell you how close you are to your “stall speed” . It’s AOA, AOA, AOA. Not airpspeed.

      • John Comley
        John Comley says:

        I feel you may have missed the essence of what is being put forward here. Your aircraft stall speed varies (increases) dramatically with increased angle of turn, which is largely what the Vdmms bug is helping you to overcome, particularly when in a low level and slowish turn.

        Please view the video on this website before dismissing Vdmms: https://av8torsafety.com/vdmms/

        Cheers, and safe skies.

  2. Joe Subits
    Joe Subits says:

    There is one perspective on the impossible turn that is almost never discussed but is the real key to deciding whether to attempt it or not. At what altitude AGL is the aircraft at when passing the other end of the runway on takeoff. This of course will depend upon runway length, surface, slope, temperature, humidity, altitude, wind and aircraft payload weight. At one end of the spectrum, a lightly loaded 180hp RV-7 with just the pilot can be at pattern altitude before it crosses the opposite end of a 7000 ft runway. At the other end of the spectrum, a 160 hp Cessna 172 loaded to gross weight would likely be at half that altitude. Maybe new operating manual charts showing safe and graveyard altitudes for each particular scenario would remove a lot of the guesswork from the equation.

    • Les Glatt.
      Les Glatt. says:

      You should read the paper, “Single-Engine Failure after Take-off: The Anatomy of a Turnback Maneuver”.
      It is on the internet. It addresses you comment on the altitude over the Departure End of the Runway (DER).
      It all about runway length and altitude over the DER.

  3. larry maynard
    larry maynard says:

    Pretty good article but do you have to use such an awkward word as taxonomy? Come on. I trust you are not trying to impress and most people don’t know what it means. If nothing else it causes the reader to pause and interrupts the flow. Best to stay away from this kind of thing.

  4. Tom Curran
    Tom Curran says:

    “Return To Runway Failed: The “impossible turn” was attempted and failed.” If I opt to attempt an “impossible turn” after takeoff, it’s based on my confidence, proficiency and experience with that airplane, under the existing conditions. “Returning to the Runway” and landing opposite direction on the pavement I just took off from is Number 3 on my Priority list. First is just getting it back over “the fence”; second is getting it down in any area clear of obstructions…a taxiway, empty parking ramp, or anywhere in the grass infield should work. I’d only attempt to land on the centerline of a runway IF I had plenty of distance & altitude to maneuver safely.

    • Frank Faludi
      Frank Faludi says:

      Good point you are making. Prior to take off if in your mind the particular field allows for getting back to a nicely cleared area as opposed to “god knows what is facing you in a more or less straight ahead direction”…well I’m more likely to make the turn back manoeuvre.

      • Tom Curran
        Tom Curran says:

        Mr. Faludi: Thank you for your response.

        My concern is that a LOT of the discussions, studies, articles, blogs, videos, Letters to the Editors…etc.…concerning the “Impossible Turn”, leave the impression that landing opposite direction on the runway you just left from is the only option. We know that is not always the case.

        Even the good folks at AOPA’s Air Safety Institute (ASI) failed to put things in that context with their much-vaunted Impossible Turn ‘flight test’ (Google it).

        The ASI staff, representing us ‘average’ pilots, flew a variety of their own airplanes, from a Super Cub to a Bonanza, through a scenario where they had engine failure on takeoff and attempted to land opposite direction on the departure runway. In each case, they either just barely made it…or didn’t even come close.

        What they didn’t point out was there were acres of unobstructed pavement and freshly mowed grass available to land on. I understand that wasn’t the objective of the “test”, but that is the reality we face every time we take off: Every airport environment presents different challenges, threats, and opportunities. Our responsibility as pilots is to have considered those before we release the brakes.

        One thing their test did reinforce, was that it is never going to be “just” a 180-degree turn if you are trying to land back on the same pavement…unless you’re doing a Split-S….and it’s often that final low-altitude, end-game maneuvering to “line up” with the runway that kills folks.

        I like the concept that if you make your initial turn “into the wind”, it will offset your turn radius and you’ll roll out on centerline. However, I’m not betting on it. Certainly, wind conditions are going to affect your landing performance; but if the direction & velocity are that significant, you should have taken that into consideration before you took off on the first place. Read the “fine print” in your POH performance tables that addresses landing with a tailwind.

  5. Mark York
    Mark York says:

    Thanks for the article. Pointing out that it does have non perfect data is honorable. As a former Army dust off pilot and civilian fixed wing pilot, having additional information at someone else’s expense is invaluable. On the notion of a no fly zone chart, we have a deadman’s curve in our UH-1V model manual. Perhaps as one commentator indicated, this could be additional life saving information. Being courageous enough to articulate information is always followed up by proof readers. (But I do welcome the proof readers as they care enough about the aviation family to support safe flying). Mark York

  6. Jim Price
    Jim Price says:

    When learning to fly my instructor (also my dad) made every attempt to point out and teach all “possibilities” that might occur. When I teach students and give flight review, I also try to open the mind so that such a surprise will not occur. Just prior to pushing up the power I always ask the PIC the “what if” question. Seems that if you do not think that there is a possibility of problems you will be surprised. Our local airport and also the one where I learned to fly, one of the main departures is over the city. Loss of power or reduced power should always be considered, and a straight ahead is not the solution. We teach so that our students will not be the ones that never experience such problems, but can handle same if they do occur. I really think practicing the Impossible Turn at altitude can open ones eyes and see that sometimes it is possible and the airplane will not stop flying if the nose is down even in a very steep turn. Practice would make this understandable, easier and give an indication of how much loss of altitude will occur during such a turn to landing. Another thing, I notice that many people in light singles and twins, seem to go directly to a cruise climb after lift off rather than Vx to achieve as much altitude prior to end of runway as possible. Fly safe and stay healthy.

  7. John T
    John T says:

    There’s no mention in either the reported small data set, nor this discussion of either a pre-brief prior to taking the runway of actions of the pilot should an engine fail (entirely or partially) on takeoff roll; after breaking ground; at 50′, at 200′, at 400′, at 600′, at pattern altitude… (or other key altitudes based on aircraft being flown, strength and direction of wind, airport configuration and size, surrounding terrain and obstacles, etc.). Turing INTO a cross wind reduces the turn radius while AWAY from the crosswind component exacerbates the return problem. Besides, when the engine quits in that critical moment the plane becomes the property of the insurance company (we do carry hull and liability, right??).

    Thanks for the article. Your data set is interesting, as were your search keywords. I hope you enjoyed your conversation with CAROL :) Perhaps adding 2017 and 2016 to your data set might be a worthwhile enterprise.

    • Ed Wischmeyer
      Ed Wischmeyer says:

      Hi John – The data contained no mention of any pre-briefings or lack there of. While those briefings would be useful in takeoff situations, I’ve never heard mention of landing briefings that included what to do in case of a botched go-around.

  8. Marc Rodstein
    Marc Rodstein says:

    A major flaw in this analysis is that it is a study of accident data, not non-accident data. A pilot who succeeds in performing the so-called “Impossible Turn” will have no accident, and that data is missing from the analysis. I have an instructor friend who fits this category. After engine failure, he sucessfully returned to the airport and landed on the runway from which he departed. Events like this do not show up in the accident records, and there is no way to know how many of these sucessful events have happened. By necessity, this fact of life skews the results by including failures in the statistics, but omitting most of the sucessful turn-back attempts.

    I have received instruction in this manuever from the aforementioned instuctor-pilot, and found it to be eye-opening. In my opinion, every pilot should get instruction in the turn-back manuever and practice it periodically at a safe altitude. Turning back is obviously not appropriate in all situations, but there are situations in which a properly prepared pilot can do it sucessfuly. An unprepared pilot, not so much. The best way to know when to consider turning back and when to reject the idea is to practice doing it. Like any other procedure, practice leads to proficiency.

    Those who categorically reject the turn-back are generally those who have never practiced it. I believe they are mistaken. The turn-back is a tool that should be in the pilot’s tool box, kept sharp, and used judiciously only when the circustances allow. But if you haven’t practiced the procedure, you won’t know when it is “possible” and when it is “impossible”.

    • Peter N. Steinmetz
      Peter N. Steinmetz says:

      I agree that a major problem here is knowing how often this maneuver is successful.

      I also had an instructor that successfully did this one time. When asked about it his comment was “what else was I going to do? The other runway was right there” he was able to turn and land on a parallel runway,

    • Ed Wischmeyer
      Ed Wischmeyer says:

      Marc, very, very few databases record events with no injuries and no damage. The ASRS is the only one that comes to mind where there are occasionally such reports. So while you are technically correct in pointing out that this analysis excluded non-accidents/non-incidents, you are in effect asking for data on unreported events.

      By the way, I have done several major studies where I studied both incidents and accidents, and discovered things that were not common knowledge. The reason is that when the study is restricted to accident data only, that is sampling bias.

      So we agree, at least theoretically…

  9. Mark S
    Mark S says:

    Ok, I know that we all want to save our airplane and the airport environment is where we like to interface with the ground. But that desire to get back can kill you if everything is not just right.

    One factor seldom considered when the impossible turn discussions occur is wind. Most of the time in GA the wind is to our nose, ergo the chosen runway. In any engine out scenario, the objective is to minimize the amount of energy (ground speed) you have at the time you interface with a fixed object. The severity of the crash is directly related the kinetic energy and the distance used to deplete that energy. The kinetic energy = 0.5 x mass (weight) x velocity squared. In other words the energy is exponentially increased with velocity.

    In many cases, the best choice is straight into the wind to a suitable crash (off-field) sight. A 10 mph head wind equates to a 20 mph difference in ground speed if that same 10 mph is on your tail.

    So let’s say you get down to 60 mph airspeed, and with headwind, 50 mph ground speed before you hit something that brings you to a stop in a quarter of a second. You would experience an immediate onset of 9.1 Gs. The same scenario at 70 mph ground speed produces almost 12.8 Gs. Now spread the crash site over half a second and the G-force goes down to 4.6 at 50 and 6.4 at 70.

    So maybe the discussion we should have is how to think about the wind in pre take off briefing and what to run into to spread out crash.

    Just a thought.

  10. Mike B.
    Mike B. says:

    Ed, I disagree with your proposal in the laast paragraph to rectify pilots ignoring a blaring stall horn. There is already too much training time spent recovering from stalls at the expense of training to avoid them. The current stall recovery training desensitizes pilots to the stall horn by dictating they continue to pull back on the yoke untill a full stall is achieved. The training time would be much better spent instictively releasing back pressure on the yoke as soon as the stall horn sounds. Being taught to avoid the stall is much more important than the emphasis on recovery. The vast majority of real stalls occur in the runway environment at altitudes low enough to make recovery unlikely anyway. Let’s quit teaching pilots to ignore the stall horn.

    • Ed Wischmeyer
      Ed Wischmeyer says:

      Mike, when flight tests for PPL transitioned from PTS to ACS, there was considerable debate (still is) on stall training, and what you advocate was in fact adopted. One point made, contrary to your argument, is that if the stall warning goes off on final when the plane is ten feet in the air, a knee-jerk reaction of pushing the wheel forward is inappropriate. This is all part of a larger theme of basic airmanship — stick and rudder skills, if you like — being neglected. This neglect is seen in private pilot applicants and in new CFIs. Aviation is too complex for a simple one size fits all situations solution.

  11. Robert Eleazer
    Robert Eleazer says:

    I think the Impossible Turn is emphasized because it represents possibly the ultimate nightmare scenario, but the problem with generating lessons learned from it is the possible variability of the situation.
    I know of the details of two Impossible Turn situations; both occurred at airports I know very very well.
    In one, the pilot was concerned that the engine did not sound right, chopped the power just after take off, and successfully made the Impossible Turn with no one injured , but with the aircraft essentially totaled and with severe but repairable damage to two parked aircraft on the ground. The aircraft was an Ercoupe, justifiably famous for its refusal to stall and spin, and that no doubt greatly influenced the decision to do that 180, aside from the fact that the alternative locations to come down were incredibly worse in that heavily built up area. But as it turned out, there was no reason to chop the power in the first place. There was nothing wrong with the engine but the throttle cable was not properly strapped down, leading to it flexing and producing a “ROoaaarrr ROoaaarrr” sound that led the pilot to conclude to chop the power. Continued flight and normal manipulation of the throttle would have led to a normal pattern and telling the A&P, who was on board, “See? That’s what I’m talking about!”
    In the other case, a pilot took off in a LongEze, felt a vibration, and heard a noise that led him to the decision to turn around. It was a good decision; landing anywhere else in the area but the airport would have destroyed the airplane and might have caused him injury. He had to handle about an 11 kt tailwind and as a result ran off the end of the runway onto a 500 ft grass area and did no serious damage to the airplane. In his case the problem was very real, not imagined. Someone on the ground had destroyed the propeller with a shotgun blast.
    So both pilots made the correct decision to return to the airport, although only one was presented with a truly Impossible situation.

  12. E Sharp
    E Sharp says:

    One thing I find missing from the IT discussions is the effect of training the IT sooner than later is a student pilot’s progression. I lucked out into having an instructor who felt that the IT should be included in a few lessons prior to solo so as a student I actually had a chance to experience just what my options might be in an actual event (and identify what others have commented on that even if you can’t make the runway the relatively clear of obstruction airport environment itself can be a very good option). They were well briefed and well conducted lessons to take a theoretical occurrence into reality… both when you have enough energy available and when you don’t.



    I wish people would stop referring to the “Possible Turn” as the “Impossible turn”.

    If you experience engine failure after take-off, there is an aircraft/altitude/airspeed/weight/pilot combination that will allow you to complete a ‘Safe Turn’ back to the departure runway.

    It was always a given that if you had a combination of the requisite items noted above, you could always return to your departure runway if you lost your engine.

    The trick was to pin down the exact combination of the five items noted above that would allow safe execution of the manoeuvre.

    The “land straight ahead or (say) thirty degrees either side of departure path came about because engine failure in the first few seconds of take-off puts the pilot in an extremely tenuous position: no speed or altitude to trade for turning performance. Any attempt to do so resulted in a broken airplane at best, and broken people at worst.

    Hence, the “never turn back” dictum.

    I view the present push to train for the “Impossible Turn” with trepidation: unless the pilot

    1. Is going to maintain currency (minimum 75-100 hours/year),
    2. Ascertain minimum altitude for procedure commencement with dead and wind milling prop (more drag with dead engine),
    3. Routinely practice the manoeuvre (which I doubt – most don’t even practice stalls/circuits),

    then I don’t hold out much hope for the success in the event of an actual EFATO.

    There’s nothing wrong with telling pilots to find the minimum altitude at which they KNOW they could execute a 210 degree turn back to the runway. Then add a fudge factor up to fifty percent for screw-ups.

    Don’t become a test pilot on your first engine failure after take-off; half-way around the turn back is not time to find out you’re out of airspeed, altitude, and ideas…


Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *